The association of olfactory dysfunction with mortality was examined in 1162 older persons without dementia or Parkinson's disease. They completed a standard 12-item test of odor identification and then were followed for a mean of 4.2 years (standard deviation [SD] = 2.6, range: 0-9) during which 321 individuals died (27.6%). The relation of olfactory score to risk of death was assessed in a series of proportional hazards models adjusted for age, sex, education, and other covariates. Olfactory scores ranged from 0 to 12 correct (mean = 9.0, SD = 2.2). In an initial analysis, risk of death decreased by about 6% for each additional odor correctly identified (hazard ratio = 0.94; 95% confidence interval: 0.90, 0.98). Thus, mortality risk was about 36% higher with a low score (6, 10th percentile) compared with a high score (11, 90th percentile). The association persisted in subsequent analyses that controlled for naming ability, disability, cerebrovascular disease, characteristic patterns of leisure activity, depressive symptoms, and apolipoprotein E genotype. The results indicate that difficulty identifying familiar odors in old age is associated with increased risk of death.